Synthetic ihc-stained digital sides generated using artificial neural networks

Abstract

Disclosed herein are systems, methods and computer-program products to create synthetic immunohistochemistry (IHC) stained digital slides or virtual tissue sections generated using artificial neural networks (ANNs). In some implementations, the created digital slides or a virtual tissue sections can be used as a ground truth to evaluate a method of analyzing IHC stained tissues.

What is claimed is: 
     
         1 . A method of generating a virtual tissue section, comprising:
 receiving, by a computing device, one or more inputs related to a desired virtual tissue section mimicking an IHC stained tissue having known values for one or more parameters; and   providing at least a portion of the one or more inputs to a generator executing on the computing device, wherein the generator creates a generated virtual tissue section corresponding to at least a portion of the one or more inputs related to the desired virtual tissue section,   wherein the generated virtual tissue section comprises is comprised of exact proportions of IHC positive stained cells and IHC negative stained cells in accordance with the one or more inputs to the generator.   
     
     
         2 . The method of  claim 1 , wherein the proportion of the IHC positive stained cells can vary between 1% and 100% of the generated virtual tissue section. 
     
     
         3 . The method of  claim 1 , wherein the one or more inputs include one or more of:
 an image;   a numbers or a proportion of the IHC positive stained cells and IHC negative stained cells;   annotation or segmentation-based inputs; and   a size, a shape and/or a location of the IHC positive stained cells and IHC negative stained cells and/or artifacts.   
     
     
         4 . The method of  claim 1 , wherein the generated virtual tissue section is a model for different tissue types including lymphoma, breast cancer, lung cancer, and prostate cancer. 
     
     
         5 . The method of  claim 4 , wherein the generator further introduces objects into the generated virtual tissue section including one or more of lymphocytes, histiocytes, blood vessels, nerve bundles, fibrotic fibers, and artifacts such as hemorrhage of necrosis. 
     
     
         6 . The method of  claim 1 , further comprising:
 performing an analysis of the generated virtual tissue section using an analysis method, wherein the analysis comprises determining an analyzed value for at least one of the one or more parameters; and   comparing the analyzed value for the at least one of the one or more parameters to the known value for the at least one of the one or more parameters, wherein the comparison is used to evaluate the analysis method.   
     
     
         7 . The method of  claim 6 , wherein the analysis method is a method for counting IHC positive stained cells and/or IHC negative stained cells. 
     
     
         8 . The method of  claim 7 , wherein the method for counting IHC positive stained cells and/or IHC negative stained cells is performed manually. 
     
     
         9 . The method of  claim 7 , wherein the method for counting IHC positive stained cells and/or IHC negative stained cells is performed automatically. 
     
     
         10 . The method of  claim 9 , wherein the method for counting IHC positive stained cells and/or IHC negative stained cells is performed automatically using software. 
     
     
         11 . The method of  claim 1 , wherein the generated virtual tissue section is used for pathology quality assurance programs for programs used to standardize breast cancer pathology, lung cancer pathology, and lymphoma pathology. 
     
     
         12 . The method of  claim 1 , wherein the generated virtual tissue section is used for testing with light microscopy and/or for testing of high resolution slide scanners. 
     
     
         13 . The method of  claim 1 , wherein the generated virtual tissue section comprises a three-dimensional (3D) phantom of tissue with the exact proportions of IHC positive stained cells and IHC negative stained cells. 
     
     
         14 . The method of  claim 13 , further comprising printing the 3D phantom of tissue using a 3D printer, wherein the 3D phantom of tissue is printed using a matrix. 
     
     
         15 . The method of  claim 14 , wherein the matrix comprises a cartridge of collagen. 
     
     
         16 . The method of  claim 14 , wherein the 3D phantom of tissue is used to standardize histology processing of tissue fixation, tissue cutting. tissue processing, IHC staining using different IHC platforms, image acquisition, and image IHC analysis. 
     
     
         17 . The method of  claim 1 , wherein generating the virtual tissue section further comprises:
 analyzing, by a discriminator, executing on the computing device, the generated virtual tissue section, wherein the discriminator either accepts or rejects the generated virtual tissue section on the analysis and wherein the discriminator uses artificial neural network (ANN) programming to analyze the generated virtual tissue section, and   wherein if the generated virtual tissue section is rejected by the discriminator, the rejected generated virtual tissue section is discarded and another generated virtual tissue section is created by the generator.   
     
     
         18 . The method of  claim 17 , wherein the ANN programming used by the discriminator comprises a convolutional neural network (CNN) based classifier. 
     
     
         19 . The method of  claim 18 , wherein the CNN based classifier comprises patchGAN or a modified version of U-Net. 
     
     
         20 . The method of  claim 1 , wherein the generated virtual tissue section is used as a ground truth to evaluate a method of analyzing virtual tissue sections.